What does Meta AI’s diplomacy-winning Cicero mean for AI? (Hint: It’s not all about scaling)
By Gary Marcus, Ernest Davis  |  Dec 13, 2022
What does Meta AI’s diplomacy-winning Cicero mean for AI? (Hint: It’s not all about scaling)
Image courtesy of and under license from Shutterstock.com
Meta’s new AI, called Cicero, is able to successfully strategize and negotiate with humans in playing the game Diplomacy, and it may have the potential to surpass other AI that play games such as Go and chess. Nevertheless, much remains unknown, and it could be a long time before its implications for AI as a whole are fully understood.


The distinguishing property of humans is to search for and to follow after truth. 

- Cicero

Cicero the human being (106-43 BC) was a celebrated politician, orator, and writer; his historical and philosophical importance is still being debated 2,000 years later. Cicero the computer program, which was announced in an article in Science on November 22, is a powerful artificial intelligence (AI) system that plays the game Diplomacy. A few days later its implications for AI are not yet clear and may not be for a long time.

Diplomacy, a complex game that requires extensive communication, has been recognized as a challenge for AI for at least the past 50 years. To win, a player must not only play strategically, but also form alliances, negotiate, persuade, threaten, and occasionally deceive. The game therefore presents challenges for AI that go far beyond those faced either by systems that play games like Go and chess, or by chatbots that engage in dialog in less complex settings.

The results themselves are, without question, genuinely impressive. Although the AI is not yet at or even near world champion level, the system was able to integrate language with game play in an online version of blitz Diplomacy, ranking within the top 10 percent of a mixed crowd of professional and amateurs, with play and language use that were natural enough that only one human player suspected it of being a bot.

Plenty of questions arise from this. How does it work? Does it have implications for other ongoing challenges in AI? Is it - as 

The content herein is subject to copyright by The Yuan. All rights reserved. The content of the services is owned or licensed to The Yuan. The copying or storing of any content for anything other than personal use is expressly prohibited without prior written permission from The Yuan, or the copyright holder identified in the copyright notice contained in the content.
Share your thoughts.
The Yuan wants to hear your voice. We welcome your on-topic commentary, critique, and expertise. All comments are moderated for civility.